Nicola Pavoni, Ofer Setty and Giovanni Violante at the National Bureau of Economic Research (NBER) attempted to determine what the optimal method - defined as raising the most people above the poverty line - of social insurance for those who are unemployed due to the cyclical nature of the economy. The results depended on how much money was allocated to the effort, as well as the skill level of whom you are trying to help.
I found a few of the insights of their conclusions interesting:
1) With a low budget, public works are far more effective than any other form of assistance. More public works project that you are taken upon, more that it helps people with the lowest skill sets (also those with the lowest income). It is a strong case of why there should be permanent public works program that pays the minimum wage.
2) Most of the social insurance money that is allocated in the US and Canada is on Job Search Assistance and Unemployment Insurance, which is only an optimal policy for the middle and high worker skill sets. As skill level correlates to income level, the policy is geared towards the middle and richer class. Our social insurance scheme is more designed to maintain current class status than it is to raise people out of poverty.